International Journal of Computational Engineering Research(IJCER)


Published on

International Journal of Computational Engineering Research (IJCER) is dedicated to protecting personal information and will make every reasonable effort to handle collected information appropriately. All information collected, as well as related requests, will be handled as carefully and efficiently as possible in accordance with IJCER standards for integrity and objectivity.

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

International Journal of Computational Engineering Research(IJCER)

  1. 1. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8 Temperature Control of Shell and Tube Heat Exchanger by Using Intelligent Controllers-Case Study 1, Mr.P.Sivakumar, 2,Dr.D.Prabhakaran , 3,Dr.T.Kannadasan 1, Chemical Engineering De pa rtm ent, M .Tech Scholar Co imbatore Institute of technology An na u niv ersity, c he n nai. 2, Chemical Engineering D ep art me nt, Associate Professor Coimbatore Institute of technology Ann a u niv ersity, c he n nai. 3, Chemical Engineering D e part m ent, Professor and Head Coimbatore Institute of technology Ann a u niv ersity, c h en n ai .Abstract Temperature control of the shell and tube heat exchanger is characteristics of nonlinear, time varying and time lag.Since the temperature control with conventional PID controller cannot meet a wide range of precision te mperature controlrequirement, the temperature control system of the shell and tube heat exchanger by combin ing fu zzy and PID controlmethods was designed in this paper. The simulation and experiments are carried out; making a comparison withconventional PID control showing that fuzzy PID strategy can efficiently improve the performance of the shell and tube heatexchanger.Keywords: Control algorith m, Fuzzy logic, PID Controller, Tuning1. Introduction In many industrial process and operations Heat exchanger is one of the simplest and important unit [1] for thetransfer of thermal energy. There are different types of heat exchangers used in industries; the shell and tube heat exchange rsystem being most common. The main purpose of exchanger is to maintain specific temperature conditions, which isachieved by controlling the exit temperature of one of the fluids (mainly hot fluid) in response to variations of the operatingconditions. The temperature control of heat exchanger is nonlinear, time varying and time delay system. For these situations,nonlinear control strategies can provide significant imp rovements over PID control [2], [12]. Control of temperature usingPID controllers, co mpared to other methods, is more effective and economical. The heat exchangers need to respond tohighly non linear features and work well under different operating points. In order to achieve a wide range of high a ccuratetemperature, neuro-fu zzy control and PID control methods were combined. The main design is to assume neuro-fuzzyreasoning control methods according to different error ‘e’ and error change ‘ec’ to get self –tuning PID parameter based onconventional PID controller. The simu lation of the controller was accomplished carrying out experiments in an actual heatexchanger system.2. Temperature Control System2.1 Principal of temperature control in shell and tube heat exchanger The control of temperature in a shell-and-tube heat exchanger is demonstrated in figure 1, with cold water flowingon the tube side and steam on the shell side [5] where steam condenses and heats the water in the tubes. The controlledvariable is the tube-side outlet temperature, and the manipulated variable is the steam flow rate on the shell-side. M cC p (Tin  Tout )  M s A (1)where, M c , M s , C p , Tin and Tout refer to cold water flow rate, steam flow rate, specific heat of water, in let watertemperature, and outlet water temperature respectively. Fig 1: Shell and Tube heat exchangerThe dynamics of the process are complex because of various nonlinearities introduced into the system. The installed valvecharacteristic of the steam may not be linear [3]. Dead-time depends on the steam and water flow rates, the location and themethod of installation of the temperature-measuring devices. To take into account the non-linearity and the dead-time, gainscheduling features and dead-time co mpensators have to be added. Also, the process is subjected to various external||Issn 2250-3005(online)|| ||December || 2012 Page 285
  2. 2. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8disturbances such as pressure fluctuations in the steam header, disturbances in the water line, and changes in the inlet stea menthalpy and so on.2.2 Mathematical Model of heat exchanger The total heat in the heat exchanger system can be exp ressed as equation 2. [5]. n Q f  Qs   Ci iVi dTi i 1 (2)where, Q f , Qs , C ,  , V and dT refer to total system heat productivity, total system heat dissipating capacity, specificheat capacity, heat transfer mediu m density, volu me, and temperature variation. Total system heat dissipating capacity Qs is given by equation 3. n Qs   ki Ai (Tin  Tout ) (3) i 1where, ki and Ai refer to heat transfer coefficient, heat transfer area of the heat exchanger system. The heat exchanger equations can be expressed as in equ ation 4. [12]Cw wqw (Two  Twi )d  C f  f q f (Tfo  Tfi )d (4)where, the subscripts w and f refer to cold and hot water of the heat exchanger system. Therefore considering all aboveequations, the differential equation of the shell and tube heat exchanger is shown in equation 5. dT  FT  N ( x   ) (5) dwhere, ki Ai F (6) Co oVThe transfer function of controlled object can be derived fro m equation 5, it is de scribed as the first-order with pure timedelay, expressed as equation 7. K  s G( s)  e (7) 1  Tswhere, T , K and  refer to time constant, system gain, and delay time.3. Intelligent Controllers3.1 PID Control of Shell and Tube Heat Exchanger PID controller is the most popular controller used because it is easy to operate and very robust. Implementation ofthe latest PID controller is based on a digital design. These digital PID include many algorith ms to improve theirperformance, such as anti wind-up, auto-tuning, adaptive, fuzzy fine-tuning and Neural Netwo rks with the basic operationsremain ing the same. The performance specifications such as rise time, overshoot, se ttling time and error; steady state can beimproved by tuning value of parameters Kp , Ki and Kd of the PID controller. The output Fig 2: Mai n Parts of Fuzzy PID Controlis mathemat ically represented as equation 8 and 9.||Issn 2250-3005(online)|| ||December || 2012 Page 286
  3. 3. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8 de(t ) 1 t de(t ) K P ty(t )  K P [e(t )  Td   e(t )dt ] y(t )  K P e(t )  K PTd   e(t )dt ] (9) dt T 0 (8) dt T 03.2 Structure and Parts of Self tuning Fuzzy Controller Fuzzy logic controller as shown in Figure 2 consists of main four parts fuzzification, rule base, inference engineand de-fuzzification [6], [8]. Fu zzy PID Self-tuning Control takes error "e" and Change-in-erro r "ec" as the input of FuzzyPID controller. Using fuzzy control rules on-line, PID parameters “Kp ", "Ki ", "Kd" are amended, which constitute a self-tuning fuzzy PID controller, the principle of which is shown in Figure 3. The language variable values of error "e" and theerror rate of change "ec" is (NB, NM, NS, ZO, PS, PM, PB) seven fuzzy values. And then setting up the suitable fuzzycontrol table for Kp, Ki, Kd three parameters tuning separately[8], [9].According to the fuzzy rules table, appropriate vagueand ambiguous methods is to be selected to make dynamic tuning for Kp , Ki , Kd . Fig 3: Structure of Fuzzy PID Control Fig 4: Input membershi p function for e and ec Fig 5: Output membership functi on for Kp , Ki , KdThe fuzzycontroller takes two inputs (e, and error change ec) and three outputs (Kp , Ki , Kd ). When the error is large, it is controlledaccording to the characteristics of PID control where the output value automatically closes to the given value. When theerror beco mes smaller to a certain extent, the fuzzy control takes effect. The input error, error change and output||Issn 2250-3005(online)|| ||December || 2012 Page 287
  4. 4. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8membership functions use triangular functions, which are shown in figure 4 and 5.Larger Kp is chosen to speed up thesystem response speed. At the same time, in order to avoid the probable differential super-saturation, smaller Ki is chosen. Inorder to avoid large overshoot, the integral is limited by setting Kd is zero. TABLE 1 THE FUZZ Y CONTROL RUL E FOR Kp TABLE 2 THE FUZZ Y CONTROL RUL E FOR Ki TABLE 3 THE FUZZ Y CONTROL RUL E FOR KdIn order to make the overshoot of the system respond relatively small and to ensure the system response speed , Kp is setsmaller, and Ki and Kd values are chosen respectively. In order to make the system have better steady state, Kp and Ki are setlarger, and to avoid oscillations near the set point, Kd is set properly. When ec is small, Kd is set middle, and when ec is||Issn 2250-3005(online)|| ||December || 2012 Page 288
  5. 5. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8large, Kd is set small. According to the given rules, the control rule table of PID parameters can be obtained and the controlrules for Kp , Kd and Ki is listed in table I, II and III.Defuzzificati on In this paper, the weighted average method to the fuzzy evaluation is used to get the precise control value withformula as shown in equation 10. n   (u ).u i i u0  i 1 n (10)   (u ) i 1 iWhere ui is the fuzzy set values,  (ui ) is membership degree of fuzzy values and u0 is evaluation result. After the threeparameters are ad justed by the fuzzy controller, the output control parameters are calcu lated fro m the equation 9.3.3 Neuro- Fuzzy Controller In the field of artificial intelligence, neuro-fu zzy refers to combinations of artificial neural networks and fuzzylogic. Neuro-fuzzy hybridization results in a hybrid intelligent system by co mbining the hu man-like reasoning style of fu zzysystems with the learning and connection structure of neural networks. The drawbacks are the complexity and the darknessof their structures. Industries use the PID technique since it is a crisp control. The self tuning of the P, I, D parameters arequite difficult and the resultant control is with overshoot and with large time constants. To avoid this a combination ofNeuro-Fu zzy PID controllers are used for controlling the temperature in the process.4. Experimental Results And Discussions4.1 Step Res ponse for the Intelligent Controllers Co mparison between conventional PID, fuzzy PID and Neuro -Fu zzy PID controller’s temperature control wasperformed. According to the analysis fuzzy controller is designed in MATLAB and the fuzzy self -tuning PID control systemmodel is designed by SIMULINK. The step response of the Fuzzy control and conventional contro l is shown in Figure 7,8.||Issn 2250-3005(online)|| ||December || 2012 Page 289
  6. 6. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8 Fig 6: Structure of Neuro-Fuzzy PID Control PID Fuzzy-PID Neuro-fuzzy Fig 7: Step Res ponse for the Intelligent Controllers Fuzzy-PID PID Neuro-fuzzy Fig 8: Response of Temperature control||Issn 2250-3005(online)|| ||December || 2012 Page 290
  7. 7. I nternational Journal Of Computational Engineering Research ( Vol. 2 Issue. 8The initial tuning of the PID controller is accomplished based on the Ziegle r-Nicholes method, and the gain coefficients areKp =1.1, Ki =0.003 and Kd =52. Fu zzy PID Controller has a small overshoot, fast response and the steady state error is lessthan 1%.4.2 Response of the Temperature control system The temperature control experiment is conducted in the actual Shell and tube heat exchanger system. The targetoutlet temperature of heat exchanger is 60˚C and figure 7 shows the response of heat exchanger.The results suggest thatneither the settling time nor control accuracy is satisfied enough when conventional PID controller is used in shell and tubeheat exchanger. The steady state error of the PID controller is greater than fuzzy PID. The experimental results shows thatfuzzy self-tuning PID control has better dynamic response and steady state error characteristics.4.3 Comparison of vari ous parameters Table IV shows the comparison of various parameters fro m the above graph for the various types of Controllers. TABLE 4 COMPARISON OF PARAMET ERS Parameters PID Fuzzy Neuro PID Fuzzy PID Rise Time 1.8 2.6 3.2 Peak over shoot 0.68 0.82 1.2 Settling Time 14 5.8 8.6 Peak Time 2.6 4.1 4.25. CONCLUSION In this paper design of a temperature control of a shell and tube heat exchanger based on Neuro -fuzzy PID controlwas discussed by comparing it with PID and Fuzzy PID. The analysis fuzzy controller is designed in MATLAB and thefuzzy self-tuning PID control system model is designed by SIMULINK. The results suggested that self-tuning parameterfuzzy PID controller has a smaller system overshoot, faster response and less steady state error thereby making it strongerthan conventional PID controller. It was thus concluded that fuzzy self-tuning PID control has better dynamic response andsteady state error characteristics. The control rule table of PID parameters was obtained and the control rules for Kp , Kd andKi was tabulated. The actual system used obtains a good control effect and can satisfy the requirements of the temperaturecontrol system of the shell and tube heat exchanger.6. References[1] H. Thal-Larsen, 1960, Dynamics of heat exchangers and their models, ASM E J. Basic Eng., pp. 489 – 504.[2] M .Chidambaram, and Y.S.N. M alleswara rao, 1992,Nonlinear Controllers for heat exchangers, J. Proc. Cont., vol 2(1), pp 17-21.[3] A.W.Alsop, and T.F.Edgar, 1998, Non-linear heat exchanger control through the use of partially linearised control variables, Chein. Eng. Commn.,vol.75, pp 155 – 170.[4] Wang Li-Xin, 1994,Adaptive fuzzy systems and control: Design and stability analysis [M], Prentice-Hall Englewood chiffs NewJersey.[5] H. Yamashita, R. Izumi, S. Yamaguchi, 1978, Analysis of the dynamic characteristics of cross -flow heat exchangers with both fluids unmixed, Bull. JSM E, vol. 21 (153), pp 479-485.[6] Hassan B.Kazemian, 2001,Development of an intelligent Fuzzy Controller, IEEE International Fuzzy Systems Conference. Vol.1, , pp. 517–520.[7] M ann G. K.I., Hu B.G.Gosine R.G, 1999, Analysis of direct fuzzy PID controller structures, IEEE Trans on Systems, M an and Cybernetics. Vo1.29, pp. 371-388.[8] M . Sugeno, 1985, Industrial applications of fuzzy control, Amsterdam, The Netherlands: Elsevier.[9] Li, H.X. & H. Gatland, 1996,Conventional fuzzy logic control and its enhancement, IEEE Transactions on Systems, M an and Cybernetics, 26(10), pp.791-797.[10] Jean-Jacques E. Slotine, Weiping Li, 1991,applied nonlinear control, Pretice-hall, Inc., London, pp. 278-282.[11] George K. I. M ann, Bao-Gang Hu, and Raymond G. Gosine, 2001,Two-level tuning of fuzzy PID controllers, IEEE transactions on systems, man, and cybernetics-part B: cybernetics, Vol. 31, No.,pp. 263-269.[12] K. M . Hangos, J. Bokor, and G. Szederkényi, 2004, Analysis and control of nonlinear process control systems, Advanced Textbooks in Control and Signal Processing, 1st Edit ion, Ch. 4, Springer-Verlag London limited, pp. 55-61.||Issn 2250-3005(online)|| ||December || 2012 Page 291